| #
d8e47b63
|
| 17-Sep-2024 |
Satish Balay <balay@mcs.anl.gov> |
Merge branch 'barry/2024-09-02/rebased-fix-conversion-warnings' into 'main'
Compiler finds (and forbid) casts from higher precision integers to lower
See merge request petsc/petsc!7806
|
| #
f2719977
|
| 09-Sep-2024 |
Barry Smith <bsmith@mcs.anl.gov> |
Viwering PetscSection displays dof as dim in the ASCII files, fix to dof
|
| #
164282b9
|
| 23-Jan-2024 |
Satish Balay <balay@mcs.anl.gov> |
Merge branch 'stefanozampini/dmplexforest' into 'main'
Updates to DMPLEX and DMFOREST
See merge request petsc/petsc!7217
|
| #
4cd8913c
|
| 15-Jan-2024 |
Stefano Zampini <stefano.zampini@gmail.com> |
PetscSection: propagate field names to global section
view names at View time, update outputs
|
| #
c574b33d
|
| 05-Dec-2023 |
Satish Balay <balay@mcs.anl.gov> |
Merge branch 'baagaard/update-plex-ex5-fault-projection' into 'main'
Fixes for projection between different dimensional meshes
See merge request petsc/petsc!7079
|
| #
67de2c8a
|
| 05-Dec-2023 |
Matthew G. Knepley <knepley@gmail.com> |
Plex ex5: Include tests projecting to fault mesh - Fix bd projection label and label completion
|
| #
57a85133
|
| 05-Sep-2023 |
Satish Balay <balay@mcs.anl.gov> |
Merge branch 'tisaac/feature-petscfe-vector-in-use' into 'main'
Use PETSCFEVECTOR throughout Petsc
See merge request petsc/petsc!6833
|
| #
2dce792e
|
| 14-Aug-2023 |
Toby Isaac <toby.isaac@gmail.com> |
PetscFE: add PETSCFEVECTOR
Reductions in the cost of computing, storing, and applying a Tabulation can be achieved for a finite element space that is a vector of "scalar" finite elements (the "scala
PetscFE: add PETSCFEVECTOR
Reductions in the cost of computing, storing, and applying a Tabulation can be achieved for a finite element space that is a vector of "scalar" finite elements (the "scalar" element itself could be vector-valued).
Reductions in the cost of storing and applying a Tabulation can't be realized until the PetscTabulation structure itself is changed, but at the very least reduction in the cost of computing a tabulation is possible with a new PetscFEType, PETSCFEVECTOR.
To complete the interface, we also add PETSCDUALSPACESUM, a dual space whose dual basis is the sum of other dual bases, that parallels PETSCSPACESUM (the space and dual space of a PETSCFEVECTOR are a PETSCSPACESUM and PETSCDUALSPACESUM, respectively).
To allow the user to specify whether the basis vector and value components of the PETSCFEVECTOR of blockwise or interleaved, we also have to make sure PETSCSPACESUM and PETSCDUALSPACESUM can have blockwise or interleaved behavior. We add both capabilities.
PETSCFEVECTOR is now used by detecting compatible vector-valued dual and primal spaces in PetscFECreateFromSpaces().
show more ...
|
| #
5fc265dd
|
| 15-Jun-2023 |
Satish Balay <balay@mcs.anl.gov> |
Merge branch 'knepley/fix-cohesive-integration' into 'main'
Plex: Fix orientation when integrating cohesive faces
See merge request petsc/petsc!6599
|
| #
81363e6f
|
| 14-Jun-2023 |
Matthew G. Knepley <knepley@gmail.com> |
Plex ex5: Update tests
|
| #
b6b5caf9
|
| 27-Mar-2023 |
Satish Balay <balay@mcs.anl.gov> |
Merge branch 'knepley/feature-hybrid-mass' into 'main'
Plex: Can now specify a mass weighting to hybrid residual
See merge request petsc/petsc!6210
|
| #
e5939c1d
|
| 20-Mar-2023 |
Matthew G. Knepley <knepley@gmail.com> |
Plex ex5: Added buried edge test
|
| #
d438e513
|
| 18-Jan-2023 |
Satish Balay <balay@mcs.anl.gov> |
Merge branch 'jed/fe-simplex-quadrature' into 'main'
PetscFE simplex quadrature
See merge request petsc/petsc!5525
|
| #
72f91173
|
| 15-Jan-2023 |
Jed Brown <jed@jedbrown.org> |
DMPlex tests/ex5: update tri and tet numerical values for prior code
This commit is in preparation for a change to the quadrature that will not affect these numerical values. I bisected on ex5_tet_0
DMPlex tests/ex5: update tri and tet numerical values for prior code
This commit is in preparation for a change to the quadrature that will not affect these numerical values. I bisected on ex5_tet_0 to find two commits that introduced numerical changes and summarize their impact here. I'm expecting the other changes to be similar.
The first numerical change was caused by
commit e1a13dae02585224ddd4368ee030f3c5e77a306b Author: Matthew Knepley <knepley@gmail.com> Date: Thu Apr 28 10:27:14 2022 -0400
Plex: Fix fault face division - Should not split a face that only touches the fault on unsplit points
src/dm/impls/plex/plexsubmesh.c | 20 +++++++++++++++++--- 1 file changed, 17 insertions(+), 3 deletions(-)
which has the numerical change:
diff --git 1/base.out 2/here.out index df7196b66a9..9534c0139ad 100644 --- 1/base.out +++ 2/here.out @@ -150,15 +150,15 @@ Vec Object: Local Solution 1 MPI process 0. 1.5 1. -0. -1. -0. 0.5 0. 0. -0.5 0. 0. +0. +1. +0. Discrete System with 2 fields cell total dim 27 total comp 6 cohesive cell @@ -222,23 +222,23 @@ Vec Object: Local Residual 1 MPI process 0. 0. 0. --3.46945e-18 -0.0833333 -0.0833333 0.0208333 -0.0625 +0.0208333 0.0416667 -0.0208333 -0.0208333 0.0416667 +0.0833333 0. +0.104167 +0.0416667 0. +-1.34017e-17 0. 0. +-9.26063e-18 0. 0. -0. -0. +7.55411e-18 0. Mat Object: Jacobian 1 MPI process type: seqaij
The second commit
commit d6177c40aeff79f5090ba5db2353273bea8a20ec Author: Toby Isaac <toby.isaac@gmail.com> Date: Wed Jun 29 10:29:10 2022 -0400
Use DMLabelStratumHasPoint() in CheckPoint_Private()
plex test ex61 shows there are corner cases where checking the labelId is necessary
src/dm/impls/plex/plex.c | 12 ++++-------- src/dm/impls/plex/tests/ex61.c | 1 - 2 files changed, 4 insertions(+), 9 deletions(-)
which had the numerical change
diff --git 1/base2.out 2/here.out index 9534c0139ad..5af4b0229e2 100644 --- 1/base2.out +++ 2/here.out @@ -142,14 +142,14 @@ Vec Object: Local Solution 1 MPI process 1. 1.5 0. -0.5 -1. -1. -1. -1. 0. -1.5 -1. +0. +0. +0. +0. +0. +0. +0. 0.5 0. 0. @@ -231,15 +231,15 @@ Vec Object: Local Residual 1 MPI process 0. 0.104167 0.0416667 -0. --1.34017e-17 -0. -0. --9.26063e-18 -0. -0. -7.55411e-18 -0. +0.0416667 +0.145833 +0.166667 +0.0833333 +0.166667 +0.166667 +0.0416667 +0.1875 +0.166667 Mat Object: Jacobian 1 MPI process type: seqaij row 0: (0, 0.) (1, 0.) (2, 0.) (3, 0.) (4, 0.) (5, 0.) (6, 0.) (7, 0.) (8, 0.) (9, 0.) (10, 0.) (11, 0.)
show more ...
|
| #
d685a852
|
| 04-May-2022 |
Satish Balay <balay@mcs.anl.gov> |
Merge branch 'knepley/fix-plex-submesh-unsplit' into 'main'
Plex: Fix fault face division
See merge request petsc/petsc!5186
|
| #
3bdf08b3
|
| 28-Apr-2022 |
Matthew G. Knepley <knepley@gmail.com> |
Plex ex5: Added buried edge test
|